I will give a higher level overview, supported by a few concrete examples, on applications of deep learning at the LHC, such as jet tagging, likelihood-free inference, anomaly detection, mitigation of data-simulation differences, and fast simulation. I will also report my experiences what are typically critical aspects for achieving impactful results by deep learning. I will also briefly discuss my view on how LHC differs with respect to many other fields and point to a few typical pitfalls in our field.
